Curing Hydrodynamic Instability During Balanced Plug Cementing

The plug cementing operation is a troublesome enterprise. It is conducted whenever a well is to be abandoned or a slot recovery operation is to be performed. Also, when kicking off a well path to drill a new well path, a cement plug is frequently placed. The cement can consist of different cementitious materials, ranging from Portland cement to other materials such as blast furnace slags. If setting a permanent plug or an abandonment plug, the placed and cured cementitious material must satisfy several requirements as described by Khalifeh and Saasen (2020) and Vrålstad et al. (2019). For convenience, the term cement is used for all cementitious materials in the following. Cement plugs should, in principle, prevent any leakage for eternity. In practice, the requirements are determined by company, local, and regional or national laws and regulations. For example, the British Columbia requirements are very different from the North Sea requirements as shown by Trudel et al. (2019). Frequently used requirements include establishing top-of-cement and pressure testing the cement plugs. If a leakage occurs, a remedial operation is normally required. Such remedial work is costly, especially offshore, as it involves waiting-on-cement and a new cementing operation. Thus, the definition of failure and success, which is used throughout this article, is straightforward. Success means no remedial work is necessary, while failure means that a remedial operation must be performed. Because of the additional hydrostatic pressure from the cement slurry, it is possible that the entire plug may move a handful of meters downward. But as long as the entire cross-sectional area is covered with cement, such a movement does not indicate failure.

Plug failures can result in a significant loss of rig time and additional material costs. In one organization, the 2023 costs resulting from plug failures have ranged from an estimated minimum of about USD 30,000, where inexpensive rigs were used, to several times that figure for offshore operations. Rig costs can be as high as USD 20,000 per hour, and the time waiting for the cement to set (usually referred to as waiting-on-cement) range from 12 to 24 hours for each attempt to set the plug. There are cases where as many as five or six attempts have been required before a plug was set properly in a particular well. The plug cementing operation in this particular well thus became extremely expensive. In addition to waiting-on-cement, time is also required for testing the plug, conditioning the wellbore, and mixing and spotting any subsequent plugs.

Because of the different regional requirements for meeting the ideal scope of plug cementing, it is difficult to describe the global success rate of plug cementing. Therefore, in the following sections, the success rate of North Sea plug cementing operations are assessed to illustrate a functioning technology. This technology is based on pumping an expandable floatable cement foundation tool. The possibility of using such a tool is supported by numerical calculations.

As will be shown, the flow from pumping cement through an open-ended pipe will very quickly change direction, and the cement will flow upward. This rapid change in flow direction indicates that a diverter tool, which leads the cement slurry perpendicularly out of a closed-ended pipe as described by Heathman (1996), a tool that is still occasionally in use, does not have any function. In principle, a balanced plug as described by Daccord et al. (2006) is feasible to place in position initially. However, a high-density fluid above a lighter fluid is not stable. The phenomenon is known as Rayleigh-Taylor instability. In principle, to be reasonably stable, the interface must be horizontal. The longer the interface is, the more unstable is this case, as shown by Saasen and Tyvand (1990) for some non-Newtonain fluids. Thus, it is difficult, or maybe impossible, to create a stable situation in a deviated well section. Observations show that it is possible to modify density differences, thickening time, and viscosity differences such that the success rate can be between 40% and 60%. Results presented earlier indicated between 25% and 70% success rate. A widely quoted number in the late 1990s for the industry average success rate in setting cement plugs was 2.4 attempts per successful plug, as presented by Calvert et al. (1995). Using a floatable cement foundation tool, this success rate increased to more than 95%, as indicated by the discussed North Sea success values.

Viscous pills have been tried as foundations for cement plugs. Khalifeh et al. (2020) showed that a viscous pill without any gel strength being formed is insufficient to hinder the flow of a high-density fluid downward in a well filled with lighter fluid. These findings fit with earlier field experience, which was verified by Khalifeh et al. (2020), that a gel pill can be used for such separation purposes. Also, a fluid pressure transmitting gel pill has been used for such purposes, as shown by Ronaes et al. (2008). Fosso et al. (2000) stated that it was possible to increase the success rate of off-bottom plugs in southern Algeria from around 25% to close to 100% by using specially designed viscous pills. The yield stresses of these pills were increased to reach close to the maximum of what the rig pumps could handle. However, preparation, pumping, and handling of such viscous and gel pills are time consuming, thus showing a need for a rapid solution.

In the case of an eccentric open-ended pipe in the hole, the displacement may occur only in the wide part of the annulus. This can also be difficult to cement in a vertical section. Both if the well sections are vertical or they are deviated, the drilling fluid is likely to move upward in the well, and the cement is likely to move downward and create a poor cementing result after placement. Most of these effects or their solutions will be demonstrated in the following sections.

A floatable cement foundation tool, described earlier by Harestad et al. (1997), was initially run on drillpipe. The present tool is pumped down the drillpipe and expanded when it exits the drillpipe. In this section, the function of the tool and how it forms a foundation for the cement is described.

The floating cement foundation tool, shown in Fig. 1 , consists of two diaphragms held by a series of ribs as shown by Daccord et al. (2006). At the bottom of the tool, there are some wiper plugs to ensure proper movement during pumping. While pumping, the diaphragms and the ribs are compressed such that the tool has the shape of a rod, like a closed umbrella. Above the upper end of the ribs, there is a release mechanism. The tool is not allowed to release the tension holding the ribs until the pads on the release mechanism are allowed to spread out to a larger diameter than the drillpipe.

The hydrostatic pressure regimes in a well with and without cement slurry are shown in Fig. 2 . The figure shows the pressure at the surface, bottom, and packer depth at half-depth for several cases. It is assumed that the pressure at the top is atmospheric and that there is no inflow into this well. Thus, the pressure will be determined by the fluid densities alone. In Case A, the well is only filled with water, and the hydrostatic pressure increases linearly with depth. If a packer is set at the packer depth, the hydrostatic pressure is trapped on both sides of the packer as shown in Case B. Because the same fluid is both above and underneath the packer, the pressure at the bottom remains constant. In Case C, the well has been displaced to a cement slurry above a packer that is anchored to the wall of the well. This cement slurry has a density of 2.0 sg, while the water below the packer has a density of 1.0 sg. Then, there will be a pressure differential from the top of the packer to underneath the packer, meaning there will be higher pressure above the packer than below because the original water hydrostatic pressure was trapped underneath the packer when the packer was set. If the packer’s anchor becomes released as shown in Case D, the pressure will equalize above the packer because the packer now will move a bit down and compress the water below. Then, the pressure differential across the packer will be zero, as the pressure above and below the packer element will be equalized.

In Well E of Fig. 2 , the packer is replaced with a floating cement foundation where the diaphragms take on the same role as the packer element and keep the fluid above separated from the fluid below. There is zero pressure differential over the diaphragms as the floating cement foundation is free to move between the water below and the cement slurry above. The density of the cement slurry above will push the diaphragms to the wall of the well and give a very good seal between the two fluids. The height of the cement slurry above the floating cement foundation does not have any impact on the force that gets applied to the diaphragms.

Numerical simulations have been used to demonstrate how the injected cement slurry may flow upward when injected into a well through an open-ended pipe. We model cement slurry as a Bingham fluid (Fluid 1) with density $ρ^1$⁠, yield stress $τ^y$⁠, and plastic viscosity $μ^1,p$⁠. Wellbore fluids are modeled as a Newtonian fluid (Fluid 2) with viscosity $μ^2$ and a lower density, $ρ^2<ρ^1$⁠. The well is approximated as a 2D vertical channel within which an open-ended channel (from here on referred to as the injector) is placed to inject the slurry. An illustration of the flow domain is presented in Fig. 3 . Note that the figure dimensions are not to scale. The width of the injector (⁠$D^i$⁠) is half the width of the channel (⁠$D^c$⁠). The top and bottom of the computational domain are sufficiently far from the injection point to ensure the flow of cement slurry is not affected by the vertical size of the domain. $e^$ is the distance between the centerline of the injector and that of the channel. The fluids are assumed to be miscible. The mixing of fluids are modeled using a phase fraction field, $ϕ=(ρ^−ρ^o)/Δρ^$⁠, governed by an advection-diffusion equation. Here, $Δρ^=ρ^1−ρ^2$ is the density difference. $ρ^o=(ρ^1+ρ^2)/2$ is the average density. The flow problem is thus governed by Cauchy’s equations of motion, continuity, and the advection-diffusion equation describing the mixing. More details of the model problem and the governing equations can be found in Ghazal and Karimfazli (2022a).

The parameters governing the flow dynamics can be grouped into six dimensionless (⁠$Π$⁠) groups. A summary is provided in Table 1 . $V^o$ is the average injection velocity. $D^$ is the diffusion coefficient, and $g^$ is the gravitational acceleration.

Fig. 4a  shows an illustrative example of flow development when the injector is centralized. Dark red (⁠$ϕ=0.5$⁠) and dark blue (⁠$ϕ=−0.5$⁠) represent the unmixed cement slurry and wellbore fluids, respectively. When injection starts, a buoyancy-driven displacement flow develops below the injector as the cement slurry invades the wellbore fluids, flowing down the channel (see $t≤9$ in Fig. 4a ). The wellbore fluids, on the other hand, flow up along the channel walls. Therefore, a multilayer simple shear flow develops below the injector on the channel walls. This unsteady flow configuration, however, is unstable (see $t=23$ in Fig. 4a ). Interfacial instabilities grow at the fluids interface, leading to disruption of the downward flow of the injected slurry.

The development of the instabilities leads to the formation of a row of vortices and enhanced mixing of the wellbore fluids and the injected slurry (see $26≤t≤90$ in Fig. 4a ). The outcome is the development of a smooth density gradient separating the wellbore fluids and the cement slurry. We refer to this fluid region as the mixing layer. The development of the mixing layer is evident in Fig. 4b , where we have illustrated the phase fraction along the symmetry line below the injector. The white dashed line indicates the time when the buoyancy-driven displacement flow is disrupted, $tb$⁠. Fig. 4b  illustrates that the mixing layer approaches a quasisteady state shortly after $tb$⁠. We thus expect that the injected cement slurry will keep accumulating along the annular space after this stage.

Fig. 5  displays an illustrative case when the injector is eccentric (⁠$e=0.1)$⁠). Fig. 5a  shows snapshots of the phase fraction field. When injection starts, the injected cement flows along the wide side of the annulus because it has lower hydraulic resistance compared with the narrow side. As the injected cement slurry flows through the annulus, a row of vortices develops below the singer. We expect that the upward flow of the injected cement along the wide side of the annulus will persist during the injection process. Fig. 5b  illustrates that the density profile below the injector becomes smoother with time. It provides further evidence that a buoyancy-driven displacement flow is unlikely to evolve below an open-ended pipe during the injection.

As the injected cement slurry appears to flow directly toward the wide side of the annulus, the fluid flowing into the narrow side of the annulus is a heterogeneous mix of the cement slurry and wellbore fluids. The cement slurry is frequently mixed with wellbore fluids before entering the narrow side of the gap. This is illustrated in Fig. 6 , where we show the development of the average phase fraction across the wide and narrow sides of the annulus (Figs. 6a and 6b , respectively). The inclined bright streaks in Fig. 6b  illustrate the intermittent changes in the quality of the fluid entering the wide side. Note that the fluid on the narrow side (Fig. 6b) is, at best, equal parts cement slurry and wellbore liquids.

Toward the end of the injection, the pipe is pulled out of the well. Ideally, as the pipe is pulled out, the accumulated slurry stays in place. It is common practice to pull out slowly to minimize disturbances to the accumulated slurry. Here, we present numerical simulations of the interface dynamics during pull-out. Fig. 7  illustrates the fluid condition when pull-out starts. The cement slurry and wellbore fluids are shown using gray and light blue colors, respectively. When pull-out starts, the accumulated slurry is assumed to be at rest with a horizontal interface coinciding with the bottom of the pipe.

In the balanced plug method, the velocity at which the pipe is pulled out (⁠$V^p$⁠) is chosen such that the cement flowing out of the pipe fills the space vacated by the pipe, $VR=V^p/V^o=1$⁠. Development of the phase fraction when $VR=1$ is illustrated in Fig. 8a . As the injector is pulled out, the interface is deformed and is no longer horizontal (see, e.g., $t=1$⁠). Like the rise of a bubble in water, the deformations then grow because the wellbore fluids are lighter than the cement slurry. The rising velocity of this buoyant plume, however, is relatively small compared to $V^p$ ; see, for example, $t=15,30$⁠, where the injector is far from the interface, but the buoyant plume is slowly growing at the interface. We note that while the disturbance of the interface due to the pull-out prompted the onset of plume formation, the growth and development of the plume are due to the density difference between the wellbore fluids and cement slurry.

As discussed earlier, the pipe may be pulled out more slowly (⁠$V^o>V^p$⁠). An illustrative example is shown in Fig. 8b , $VR=0.5$⁠. Similar to the previous case (⁠$VR=1$⁠), the interface is deformed as the injector is pulled out. The deformation, however, grows more rapidly here. A buoyant plume escapes the interface and invades the injector (see, e.g., $t=5$⁠, $15$⁠). Another buoyant plume then starts to evolve at the interface, suggesting that the formation of the buoyant plumes may be periodic. We hypothesize that without a mechanical obstacle protecting the interface, the periodic development of buoyant plumes will stop only when the cement starts to harden.

The numerical study shows clearly that placing cement using a balanced plug technique is feasible. No diverter tools should be necessary. These have little impact on the displacement process itself. Statistical evidence shows that 40–60% of these open-ended cement plug operations fail. Based on the numerical analysis, failure happens primarily after the placement and not during the placement. The cause of the failures must be the Rayleigh-Taylor instability rising from having a high-density fluid on top of a low-density fluid. Thus, the floating cement foundation tool hinders this instability circulation. This is elaborated thoroughly in the section about interface dynamics during pull-out.

Another practical result appearing from the numerical study is that there will not be any strong cement jet streams down into the well that will destroy the diaphragms of the foundation tool. Neither should the cement flow during placement be able to misplace this tool or rotate it even though the tool can move freely up and down into the well.

Floating cement foundation tools have been applied since 2002 and more than 4,000 of these tools have been run globally to date. Little information exists in open forums for such applications. Thus, the different operators plan their jobs based on their internal statistics. However, some statistical data exist from the North Sea region. Together with good cementing practices, around 800 tools have been run by one cementing company in this region. Only 31 cases of these 800 were documented to be unsuccessful. This gives a success rate greater than 96%, which is considerably higher than the industry average of 40–60%. As was mentioned earlier about the phenomenon of Rayleigh-Taylor instability, larger hole sizes and deviated sections are more unstable after placement than smaller sizes and vertical sections. Hence, a success rate as high as 60% seems unrealistic for cementing without support for these sections.

• More than 100 cases had a well inclination above 70.

• More than 350 of the cases were in a 12 ¼ in. hole size or larger.

• More than 300 of the cases used the tool for kick-off plugs.

In some cases, a viscous fluid was used below the floating cement foundation tool. The success rate was only marginally improved, if at all. Recently, the floating cement foundation tool has been used more and more for plug and abandonment work and modified tools have been pumped through production tubing that also include geometrically challenging equipment such as side pocket mandrels. Still, the success rate seems to be unchanged.

In the Gulf of Mexico, the floating cement foundation tool was used as a foundation for a cement plug that should stop the leak of a 16-in. casing shoe. After drilling out the cement from the primary cementing job remaining in the 16-in. casing shoe, set at 11,585 ft depth, drilling continued to 13,500 ft with a 20-in. under-reamer tool. Assessment of the leakoff test performed on the 16-in. shoe area indicated that the cement shoe was leaking. Hence, the leakoff test was not accepted, and it was decided to set a cement plug and squeeze the shoe area. A floating cement foundation tool was set 30 ft below the 16-in. casing shoe in the 20-in. underreamed hole as illustrated in Fig. 9 . The drilling fluid density in the well was 1.35 sg, and the cement slurry density for the squeeze operation was 1.97 sg. The cement slurry was placed in the well above the floating cement foundation tool, and a squeeze operation was performed by squeezing 14.3 m³ into the 16-in. casing shoe area. After the cement had set, the cement was drilled out, including the floating cement foundation tool. Below the floating cement foundation tool, no cement was found. This lack of cement indicates a fully working floating cement foundation tool. Without the floating cement foundation tool, it would be impossible not to find any cement below the depth of the tool.

Numerical analysis of cement flow and assessment of plug cementing results show the following:

• Balanced plug placement of cement is feasible.

• The cement flow out of an open-ended pipe changes direction very quickly, thus there is no need for any diverter tools in plug cementing.

• Pulling the pipe out of the cement may destabilize the cement plug.

• The success rate of conducting a balanced plug operation without any support underneath the open-ended pipe is 40–60%.

• Experience shows that when a floating cement foundation tool was used in North Sea balanced plug operations, the success rate increased to more than 96%.

The authors thank Perigon for allowing us to publish this work. The numerical studies have been carried out at Concordia University. Ida Karimfazli gratefully acknowledges the financial support provided by PTAC-AUPRF via grant PTAC-17-WARI-02 and from NSERC via grant CRDPJ 516022-17 (“Plug and Abandon Strategies for Canada’s Oil & Gas Wells”). Abdallah Ghazal acknowledges the support of Le Fonds de recherche du Québec–Nature et technologies (FRQNT). This research was enabled in part by support provided by Calcul Québec (http://www.calculquebec.ca) and Compute Canada (http://www.computecanada.ca).

This paper (SPE 214577) was accepted for presentation at the SPE/IADC Middle East Drilling Technology Conference and Exhibition, Abu Dhabi, UAE, 23–25 May 2023, and revised for publication. Original manuscript received for review 18 March 2023. Revised manuscript received for review 6 May 2023. Paper peer approved 10 May 2023.